Tree-Based Methods

Section 6.1 covers standard methods for creating decision trees. Section 6.2 applies a commercial version of Ross Quinlan’s C4.5 decision tree algorithm to customer churn data. Section 6.3 presents rpart, a tree-based learning model for constructing decision and regression trees. Section 6.4 introduces the RWeka machine learning package and applies its J48 decision tree algorithm to customer churn data. The focus of Section 6.5 is on ensemble techniques including bagging, boosting, and random forests, for improving performance. Building regression trees with rpart is the topic of the final section of this chapter. A total of eight scripts and several end-of-chapter exercises provide experiential insight into tree-based methods.